Fix unsafe indexing on dataframe and remaining unit tests

Author: PGijsbers

openml/datasets/functions.py

@@ -267,9 +267,10 @@ def check_datasets_active(
         A dictionary with items {did: bool}
     """
     datasets = list_datasets(status="all", data_id=dataset_ids, output_format="dataframe")
-    missing = set(dataset_ids) - set(datasets["did"])
+    missing = set(dataset_ids) - set(datasets.get("did", []))
     if raise_error_if_not_exist and missing:
-        raise ValueError(f"Could not find dataset(s) {missing} in OpenML dataset list.")
+        missing_str = ", ".join(str(did) for did in missing)
+        raise ValueError(f"Could not find dataset(s) {missing_str} in OpenML dataset list.")
     return dict(datasets["status"] == "active")
 
 

tests/test_datasets/test_dataset_functions.py

@@ -140,7 +140,7 @@ def test_check_datasets_active(self):
         self.assertIsNone(active.get(79))
         self.assertRaisesRegex(
             ValueError,
-            "Could not find dataset 79 in OpenML dataset list.",
+            r"Could not find dataset\(s\) 79 in OpenML dataset list.",
             openml.datasets.check_datasets_active,
             [79],
         )

tests/test_runs/test_run_functions.py

@@ -1414,7 +1414,7 @@ def test_get_runs_list_by_uploader(self):
 
         runs = openml.runs.list_runs(uploader=uploader_ids, output_format="dataframe")
         self.assertGreaterEqual(len(runs), num_runs + 1)
-        for run in runs.to_dict(orient="index"):
+        for run in runs.to_dict(orient="index").values():
             self.assertIn(run["uploader"], uploader_ids)
             self._check_run(run)